Learning neural network weights using genetic algorithms-improving performance by search-space reduction

The authors present a technique for reducing the search-space of the genetic algorithm (GA) to improve its performance in searching for the globally optimal set of connection-weights. They use the notion of equivalent solutions in the search space, and include in the reduced search-space only one solution, called the base solution, from each set of equivalent solutions. The iteration of the GA consists of an additional step where the solutions are mapped to the respective base solutions. Experiments were conducted to compare the performance of the GAs with and without search-space reduction. The experimental results are presented and discussed